Homelab Project - Proxmox High Availability Cluster with Ceph
ฝัง
- เผยแพร่เมื่อ 23 พ.ค. 2024
- Want to have some fun with your Homelab? This tutorial will take you through setting up a high availability Proxmox cluster, show VM migration using basic storage and then configure a Ceph cluster for more advanced failover.
0:00 Intro and server installation
1:35 Cluster configuration
6:23 Simple VM migration
8:50 Breaking things
14:10 Ceph configuration
24:11 Migration with Ceph
29:54 Final thoughts
#homelab #proxmox - วิทยาศาสตร์และเทคโนโลยี
Proxmoxception, I love it
At 28:30 - There is a setting under Datacenter > Options > HA Settings. Set the policy to Migrate. The default is shutdown. This way when you shutdown a host, it will live migrate things first that are in HA groups, rather than shutting them down.
Thank you!
💖💖💖💖
Really cool to see! Are you able to live migrate LXCs using CEPH?
I think you can but I haven't tried it. I'll need to check that out.
Live migration with LXC is not a thing in proxmox (or LXC in general). It's not a VM. The container is stopped, then started on the other node. There is a restart process for the container involved.
If you don't want to use CEPH you can use ZFS with replication which will give you HA. The data replication won't be in real time but you can configure that per VM or CT. Smallest is 5 minutes for replication. Some of my VMs are mostly static or little data changes I set them to replicate every couple of hours and it's usually pretty quick.
I've used CEPH in production at work and later switched to ZFS with replication to keep things simple for me as don't have time to troubleshoot issues. As a CEPH cluster if something should go wrong it could effect the entire cluster. Where as with ZFS it only affects the node. Newer version of CEPH has come a long way since I last used it but I prefer ZFS.
I will check that out. I would love to have a proxmox cluster double as a simple nas so I could pool all of my storage across multiple machines. I think ceph could do that but it isn’t worth then hassle.
Hello I have the same issue in setting up as you did where the VM migrated to a host without its disk vol now I’m stuck. How do I migrate or configure the vm back to its orig host with its data. Thanks for the video.
I’ll need to check on that. I couldn’t find a way to do it using the web ui.. might have to use the command line.
I think you create replicated pool so you are loosing a lot of storage space ... why do not create erasure coding pool ?
Honestly was just being lazy. Since I was building it to learn and it was using layers of hypervisors I didn't put any effort into configurating the pool well.
replication is computationally more efficent. In a prod environment, you want seperate nodes just doing ceph, with your vm nodes handing the compute.